1. Introduction. Brain metastases of breast cancer are thought to increasing in incidence, particuarly among metastatic patients with Her-2+ or triple negative tumors, and confer a dismal prognosis. Our goal is to understand the nature of the blood-tumor barrier (BTB), the endothelial-associated structure that results when the formation of a metastasis disturbs the normal blood-brain barrier (BBB). In initial studies the paracellular permeability of the BTB was tested in experimental brain metastasis model systems in mice, using dyes and drugs. Published collaborative research showed that experimental brain metastases of breast cancer were heterogeneous in their permeability, both within and between metastases in the same brain. While most metastases were permeable as compared to the normal BBB, only 10% exhibited sufficient permeability to enable a cytotoxic response to a systemic drug. 2. Cellular and protein characterization of the BTB. Quantitative immunofluorescence of the BTB in three model systems was performed. The relative expression of known BBB components was compared between normal (uninvolved) brain and brain metastases, and also between brain metastases that were either poorly or highly permeable to Texas red dextran. Many consistent alterations were observed between normal brain and the BTB, including endothelial cell size, neuro-inflammation, absence of astrycyte endfeet, increased VEGF, etc. In contrast, few differences were observed between poorly and highly permeable metastases. The relative expression of subpopulations of pericytes was the major difference observed. Highly permeable lesions exhibited an increase in Desmin+ pericytes and a decrease in CD13+ pericytes. This work presents the first description of the BTB. Desmin+ pericytes are one of the first markers of higher brain metastasis permeability. Current research aims to further characterize Desmin+ pericytes and determine their functional contribution to BTB permeability. 3. Molecular characterization of BTB permeability. A series of experiments identified molecular alterations correlated with BTB permeability using gene expression analysis. Permeable and impermeable brain metastases were laser capture microdissected from mouse brains and both human (tumor cell) and mouse (brain microenvironment) gene expression determined on microarrays. Most of the gene expression changes correlated with metastasis permeability were from the microenvironment, rather than the tumor cells. The sphingosine-1-phosphate receptor 3 (S1P3) was identified as overexpressed in more permeable metastases. This trend was confirmed at the protein level, and S1P3 expression was localized to astrocytes in the neuro-inflammatory response. Using a S1P3 antagonist and S1PR3 gene knockdown, in vitro TEER assays (modeling the BBB and BTB) have shown that reduced S1P3 expression and function causes altered BTB permeability. In vivo a S1P3 antagonist functionally modulated the permeability of the BTB. A manuscript is in revision. 4. We published experimental brain metastasis data indicating that temozolomide, a first line treatment for primary brain tumors, was 100% effective at preventing brain metastases of 231-BR cells over a two log dose response. This type of profound prevention has not been previously observed. Given later when brain metastases are already at least partially established, temozolomide was ineffective. Activity was dependent on methylguanine methyltransferase (MGMT). Staining of matched sets of primary breast cancers and resected brain metastases showed poor concordance, but 60% of the brain metastases were low in MGMT. A phase II trial has been developed in the WMB to test the effect of TMZ on preventing brain metastases in patients with 1-3 brain metastases from HER2+ metastatic breast cancer, having only local brain metastasis treatment (SRS or neurosurgery). Patients will be randomized to T-DM1 or T-DM1 + temozolomide, with a primary endpoint of oneyear freedom from a new brain metastasis.